Channel prediction for underwater MIMO communications using adaptive bidirectional gated recurring unit
|
Full Text |
Pdf
|
|
Author |
Hitesh D., Saiteja V. and Ravikumar C. V.
|
|
e-ISSN |
1819-6608 |
|
On Pages
|
904-911
|
|
Volume No. |
19
|
|
Issue No. |
14
|
|
Issue Date |
October 12, 2024
|
|
DOI |
https://doi.org/10.59018/072421
|
|
Keywords |
adaptive bidirectional gated recurrent unit, multiple input multiple output (MIMO), internet of underwater things (IoUT).
|
Abstract
As the realm of the Internet of Things (IoT) continues to evolve, niche applications such as underwater communication are gaining momentum both in academic and industrial spheres. Within this context, the utilization of multiple-input multiple-output (MIMO) technology holds immense importance for bolstering channel capacity in underwater acoustic (UWA) communication setups. Accurately forecasting channel responses emerges as a critical aspect for ensuring optimal system functionality. This paper introduces a streamlined model for predicting channel impulse responses (CIRs) tailored specifically for UWA MIMO communication scenarios. Dubbed the small adaptive bidirectional gated recurrent unit (ABiGRU) network, our model exhibits the ability to discern channel characteristics without necessitating intricate knowledge of internal channel properties. The proposed approach leverages short-term CIR data for real-time training, subsequently enabling accurate predictions to track the dynamic nature of UWA channels. To validate our methodology, we integrate space-time block coding (STBC) with minimum mean square error (MMSE) pre-equalization within the UWA MIMO framework. Our simulations demonstrate the practicality of this scheme, showcasing low bit-error rates (BER). Furthermore, we conduct an extensive evaluation of our ABiGRU network's prediction accuracy vis-a-vis the widely employed MMSE algorithm and other recurrent neural network (RNN) variants like gated recurrent units (GRU) and long short-term memory (LSTM). Real-world experiments in UWA MIMO settings underscore the superior performance of our ABiGRU network, suggesting its potential for cost-efficient deployment in underwater IoT sensor networks.
Back